Talk:System bus model

PROD
This is a machine architecture, according to the diagram, and according to various books. It needs to be rewritten, to distinguish that this is not a CPU architecture. (hence the CPU is a unit in this model, not the entirety of the model). I/O devices, main memory and the CPU all exist on a system bus to communicate with one another, the system bus is composed of data lines, address lines and control lines. That is the standard way that PCs are built on when the system bus was a local bus (ie. the original IBM PC, when simplified to a basic diagram)

76.66.200.95 (talk) 04:50, 19 October 2010 (UTC)


 * So what the article describes is wrong, thus efforts to find references supporting its notability are also flawed since the search terms used do not accurately summarise the topic's essentials? Rilak (talk) 04:44, 20 October 2010 (UTC)


 * I left a note about it at Wikipedia_talk:WikiProject_Computing, and someone has already said that deletion is not uncontroversial. What the article describes is a model where the CPU is a concept that exists, whereas in the original VNA, the concept of the CPU does not exist, since the VNA predates the CPU. This article is about a computer architecture, that includes the concept of a CPU, and is in many different books, called derivative of the Von Neumann architecture, and in other books, is classified as a Von Neumann architecture. The description, as sparse as it is, is meaningfully accurate except for its statement about combining the ALU with the CPU, which is wrong. It combines the ALU and the CU (control unit). 76.66.200.95 (talk) 10:51, 20 October 2010 (UTC)


 * IIRC, von Nuemann proposed a computer system comprised of five? "parts" which he called organs: "central arithmetic", "central control", memory, input, and output. What does this have to do in addressing the notability of the system bus model? Nothing, just like your comment. Rilak (talk) 08:35, 21 October 2010 (UTC)


 * I suggest that further discussion also occur at WikiProject Computing, as it will more likely gather other interested editors. 76.66.200.95 (talk) 10:51, 20 October 2010 (UTC)


 * Discussion about this article should occur here. --Kvng (talk) 13:09, 20 October 2010 (UTC)
 * IMHO: unless I'm missing something where it is already described, it should stay (such concept is mentioned, even if not by this name, in every second book on computer architecture), and should be cross-referenced with Symmetric multiprocessing and Front-side bus (both are implementations of this model, even if in wide sense). Ipsign (talk) 05:11, 21 October 2010 (UTC)

I fail to understand why you are explaining to me what the von Neumann architecture is. My issue with this article is WP:N. My issue is not with this article is not with its content (I don't care what it says, I care about how notable it is). The editor who removed the PROD claimed that there were many references of the topic, which is the system bus model. If this is the case, then why did my searches return a mere handful of results? I expect to be presented with references or with the search terms used to produce the reported results. Rilak (talk) 06:21, 21 October 2010 (UTC)


 * FWIW, as I've stated above: I support keeping this article, and there is clearly no consensus on deletion; per WP:DEL, anybody disagreeing can simply remove the PROD tag, which has been done by 76.66.200.95 (and which I would do myself if I'd run into it). In addition, I don't see how argument "why did my searches return a mere handful of results?" can possibly be relevant to article being not notable; as I see it, article can be proven notable via references, but not vice versa. In addition, it can easily be a case of notable subject but not notable article title here (which may qualify it for renaming instead of deletion). What I know (personally) is that such concept does exist (and how it is named, is another story which I don't really want to be involved into); on the other hand, proving its notability is beyond my interests right now; what I know is that there is clearly no consensus for deletion. Ipsign (talk) 07:09, 21 October 2010 (UTC)


 * Sigh. I give up. Present evidence that the term "system bus model" is notable else the article will go to AfD. If the "system bus model" is an obscure synonym, then redirect/move/merge it somewhere. Rilak (talk) 08:35, 21 October 2010 (UTC)
 * I hope author will provide evidence, if not - see you on AfD discussion. BTW, I wouldn't object to renaming this article or to merging it (for example, into an article on von Neumann architecture); I have no idea if it is worth its own article, but feel that somewhere on Wiki this concept should be described, and therefore that deleting this information completely wouldn't be a good thing. Ipsign (talk) 09:19, 22 October 2010 (UTC)


 * If Von Neumann architecture is the right destination, why don't we put up some merge banners and avoid AfD? --Kvng (talk) 12:07, 22 October 2010 (UTC)


 * I don't feel like proposing such merge myself (I'm not 100% sure about it), but if you propose it, I will not object. Ipsign (talk) 10:36, 24 October 2010 (UTC)


 * I went a little deeper, found a second ref. Merge appears to be appropriate. I've put up merge banners. --Kvng (talk) 14:44, 24 October 2010 (UTC)
 * No objections from my side. Ipsign (talk) 06:29, 25 October 2010 (UTC)

Revisited
I just realized that since there is no "I/O bus" on the diagram, if the architectural concept of Memory-mapped I/O was confused with the implementation technique of a system bus. For example, the PDP-8 could be implmented with separate buses (as in early models) or the system bus (Omnibus) of the PDP-8/E. It had explicit I/O instructions. The PDP-11 had no I/O instrcutions, so all I/O was memory mapped. It could also be implemented as a single system bus (the PDP-11/20) or separate I/O and memory buses (PDP-11/70 etc.). W Nowicki (talk) 20:49, 25 May 2011 (UTC)

Merge to System bus
With page System bus recently created by W Nowicki, I think the best we can do now is to merge System bus model into System bus (they address the same concept and are heavily redundant). Ipsign (talk) 06:11, 26 May 2011 (UTC)


 * Three questions for anyone reading this discussion:

The system bus model is claimed to be, by the article and the references it cites, an evolution of the von Neumann architecture where the CPU, memory, and I/O are connected by a system bus. The system bus is a bus that the CPU, memory, and I/O are connected to.
 * Is the following true or false?
 * Is the following true or false?
 * If the answer to questions #1 and #2 is "true", then is the system bus model and the system bus be the same concept?


 * My answers to question #1 and #2 is "true"; and to question #3 is "no". Therefore, I oppose merging system bus model into system bus on the basis that they are the same concept; because they are not. Rilak (talk) 08:48, 28 May 2011 (UTC)


 * Regarding the questions above; I have since edited the article for quality (I can't resist improving articles). The concerns my edits addressed were discussed in the AfD discussion and were not for the purpose of skewing the reaction in my favor. In the interests of transparency, here is the diff of my changes . For a full explanation, please consult the article history. Rilak (talk) 09:16, 28 May 2011 (UTC)

One nit: What von Neumann I think called "Central Control part" and "Central Arithmetical part" were what is now known as the CPU. At least according to http://qss.stanford.edu/~godfrey/vonNeumann/edv-an.pdf and http://qss.stanford.edu/~godfrey/vonNeumann/vnedvac.pdf.

Your other edits are reasonable, but do not totally follow your conclusion. Certainly there is a valid but subtle distinction between "a model that uses X" and "X" itself. But the question is if the topic is independently notable enough to merit a stand-alone article. The comments from October 2010 argue that only one (or maybe two) authors consider this a "model". Both those discussions are just short mentions in larger sections that discuss the system bus itself. I could not find any academic paper, for example, that discusses this as a "model" except the ones noted in the AfD where they are are talking about constructing a statistical model of a system bus, different concept.

And I am still confused if the author is talking about "achitecture" in the computer science sense (something visible to programer), which would be memory mapped I/O which removes the I/O bus? Or the modularity implementation technique? For example, the Omnibus of the PDP-8/E had a system bus but not memory mapped I/O. For that matter, it seems perfectly possible to implement a von Neumann machine with a system bus (binary compatible with the IAS machine say, but modern technology). In theory at least. And combining control with ALU is just implementation; they are still there conceptually just can all be on the same chip.

Certainly as a reader, I find orphan articles on subjects closely related to others with only one or two sources not as useful as a merged article, although that is somewhat taste. How about one or two sentences that say "some computer scientists consider this a streamling of the von Neumann architecture" and add the updated Murdocca book to the system bus article?

I do not see many precedents where there are separate articles on "X" and "X model". For example, there is System-on-a-chip but not a separate one on computers without any external bus, which might be a "model". There is one article on microcode, not a separate one on microcode model for the model of computers that use microcode, etc. von Neumann himself of course deserves his won article since many publications discuss him independently of his computer work. W Nowicki (talk) 22:56, 28 May 2011 (UTC)


 * Regarding the ALU being combined with the CPU, I know it's "wrong", but that's what citation says about how the system bus model differs from the von Neumann architecture (or so I assume, since I don't have access to the source cited). If I were to "correct it", I would be doing improper synthesis because I would have taken two viewpoints and reconciled them to my satisfaction by making a new conclusion that is not supported by any source. The only solution to this problem is to delete the article.


 * To clarify; my comment was to avoid my edits to the article from being misunderstood as attempts to revise the article from being about the system bus to the system bus model. Evidently; AfD and discussions elsewhere frequently do not include enough scrutiny of arguments nor the content in question. The last thing I want is for an editor to hint at something and then have ten or so other editors merely echoing that hint.


 * Regarding what "architecture" refers to in the context of this topic, I think it refers to computer organization, since the von Neumann architecture from the perspective of the programmer does not exclude memory and I/O from being connected to the processor by other means such as a crossbar switch. But I am not completely certain, since the article and the book it cites talks about combining the control with arithmetic as an organizational concept, which as you have said, is an implementation issue.


 * Looking at books and lecture notes on the Web, these materials state that von Neumann architecture consists of a CPU containing control, arithmetic, and registers, not what the system bus model claims. (Although I am aware that the First Draft states that a computer consists of, IIRC, central control, central arithmetic, memory, I/O, and an accumulator; so this will have to be discussed).


 * Regarding mentioning the system bus model in the von Neumann architecture article, I don't think it is warranted. There simply isn't enough coverage of it for it to be considered a "notable" minority viewpoint, so including coverage of it elsewhere would be undue weight. I mentioned this in the AfD, but it was mostly ignored. There really isn't a need for all documented viewpoints to be covered in Wikipedia. We simply cannot establish whether a "viewpoint" is an actual alternative viewpoint, a lie to children, or a mistake. In the case of the system bus model, I think it is the second case, since it is only found in introductory textbooks and no where else (this was raised in the AfD too).


 * Finally, I don't think that we need "models of x" articles unless there is significant coverage of it; or even a mention of it in a broader article unless the model is unique to that topic (there are going to be generic models that can be applied to many things). In this regard, "system bus model" is just a coincidental juxtaposition of "system bus" and "model"; not worthy even as a redirect (in case it is argued that deletion is not warranted, or that a legitimate article can be written under this article's title). Rilak (talk) 01:49, 29 May 2011 (UTC)

The problem, of course, is that with consensus for neither a delete nor merge, the article stays in wikipedia. To be clear, I do not think at this point any content from this article should be merged. Just change it to a redirect. The reasoning is that this can be done relatively quickly (without waiting for an administrator at least), and if anyone in the future wants to recreate it based on the one or two users of this term, they would see this dicussion and with any luck our reasoning would convince them not to. The alternative would be to wait until someone has the fortitude to propose deletion again. It seems to be an orphan still so perhaps not critical.

And I do not think it is synthesis at all to conclude that if the vast number of sources say one thing, a single source that says something else is not as reliable, so does not deserve presenting its viewpoint. Except perhaps to mention it in passing in another article, as I did in system bus, but maybe Murdocca Heuring should just be dropped from that too? Would not object to that either. W Nowicki (talk) 20:05, 30 May 2011 (UTC)

To refute a couple statements in the "speedy" denial of the second deletion: there should be no "red link" issue, since the article is an orphan. The only links are from talk pages (which the change to a redirect might help, but this should be a minor issue). There should also be no search issue. The first page presented on such a seach would be system bus if we use redirect or delete, so nothing to be gained by keeping. I also ran across Bus Functional Model (probably should not be in capitals, and has no citations, sigh) which covers the other usages of "system bus model": a statistical model of a system bus. I went ahead and changed the complaintemplates to match my thinking. The real issue is notability of this model for needing its own article, not that there is any content that should be merged. I did not go all the way and change to a redirect because those tend to be forgotten. W Nowicki (talk) 18:09, 1 June 2011 (UTC)
 * FWIW: I was one of those who opposed deletion of System bus model earlier, but with current state of affairs (after IMHO great work of User:W Nowicki on System bus), I would certainly support deleting System bus model (or redirecting from System bus model to System bus) now. Ipsign (talk) 14:54, 3 June 2011 (UTC)

In case there's anyone thinking of redirecting "system bus model" to "bus functional model", I'm not sure whether the former is a "valid" synonym for the latter. Although I mentioned in the first AfD that there are examples where the former "means" the latter, there are no sources that I know of that states this. I interpreted "system bus model" to refer to a model of a system bus based on the context of its usage. I think this was a reasonable action, since my intention then was to say that not all instances of system bus model were referring to this article's topic and not that the terms are synonymous according to sources. Rilak (talk) 22:32, 1 June 2011 (UTC)


 * My 2 cents is that this at best should be mentioned somewhere and ideally would be deleted. It's just not a useful or even vaguely commonly used term. I am one of those people who see little harm in having an article on every vaguely notable band but do see harm in documenting non-standard terms in a way that indicates they are standard/commonly used. Maybe we should have an article on the way-too-many different names people give for different bus architectures (front side bus, back side bus, local bus, split-transaction bus, memory bus, peripheral bus, serial bus, etc.  This could be mentioned there.  Hobit (talk) 03:44, 4 June 2011 (UTC)
 * Given the comments above and the improvements of System bus I'm going to boldly do a redirect to there. If you undo the redirect, please discuss... Hobit (talk) 19:52, 4 June 2011 (UTC)
 * I object to redirecting this article. The redirect is misleading, it's inaccurate, and it's not verifiable. This article should have been deleted for the reasons presented at the AfD, reasons that have not been shown to be illogical. Instead of commenting on those reasons, the AfD was derailed instead to force coverage of the system bus to be written. While I acknowledge that the redirect was done in good faith to render a quick close to this matter, I don't think it is a satisfactory solution since it creates new problems. Rilak (talk) 09:04, 7 June 2011 (UTC)


 * Glad to see this sorted out. --Kvng (talk) 14:57, 6 June 2011 (UTC)

Well evidently not sorted out yet! I do see issues with either solution: a delete removes this discussion for example, so might make it more likely for it to be created again. On the other hand, if the deletion discussion is kept, that should suffice. So if Rilak proposes a deletion, I would support. Cerainly I never intended to "derail" anything and not sure anyone did: just wanted to add to the content. Can we do a Prod on a redirect? That might be the quickest, but seems not allowed. I should also note that an anonymous IP 131.211.84.85 editor did atempt to merge in some of the content, which I do not necessarily agree with. A system bus could be multiplexed, for example. In general that article was meant to just be a start, and review by other editors would be appreciated. W Nowicki (talk) 15:52, 7 June 2011 (UTC)


 * I did not intend to say that you derailed the AfD, but I did get the impression that some editors were far more concerned with the fact that the system bus model article contained content that could have been used elsewhere than the serious problems it had; and refused to discuss the AfD accordingly. I suppose it's a bit like the reverse of throwing out the baby with the bathwater: keeping this article's content because of the lack of a system bus article, even though that was a separate issue that could have been resolved separately. Now we are stuck with a consensus and a bureaucratic perspective on the matter &mdash; refusal to reconsider the decision for an arbitrary length of time regardless of anything.


 * Regarding deleting the redirect; the correct venue to raise this is Redirects for discussion, but given that it has only been a week or so since the second AfD, it will not be well received.


 * Regarding the recent merge of content from this article to that about the system bus, I think it should be reverted. The merged content is out of place: it does not consider what the rest of the article is saying. It is more than a stylistic issue; it is like that content can never be any more than what it is because it was summarized from an introductory text that was meant to convey a specific point and not something more general. For instance, it was pointed out that the address and data buses could be multiplexed. The Q22 bus for example. I also raised this issue in the first AfD in response to the initial merge positions. Rilak (talk) 05:42, 8 June 2011 (UTC)

Yes. What bothered me also was that it had a link to data bus which went back to itself! Since generally any term that redirects should be mentioned in the lead in bold, I added back in one sentence about the different functions (which may or may not be separate pysically as we know). Evidently address bus and control bus have their own articles, albeit not well cited ones. And then there is expansion bus which might be redundant, and not cited either.

My guess is you were talking about the Q-Bus, not the public transit line in Queens? :-) I will add it to the disambig page. It is notable enough to be on computer bus, but it looks like Unibus is missing and VAXBI Bus - need to fix that too. W Nowicki (talk) 20:35, 8 June 2011 (UTC)


 * Yes, I meant Q-Bus, which IIRC was also called Q22. Looking at diffs at system bus it looks like that the merged content was replaced, correct? That should get rid of any attribution issues that might prevent deleting the system bus model redirect in the future. And looking at Wikipedia's set of articles about buses; it's a mess. Data bus redirects to system bus; therefore buses that are not system buses don't have paths for data?! And Computer bus, instead of being about generic universal bus properties, it misses the point and is redundant with Expansion bus. There needs to be a discussion about how to fix this somewhere. Rilak (talk) 00:36, 9 June 2011 (UTC)

File:Computer system bus.png Nominated for speedy Deletion
"you" should delete it, since I just forgot to update this one to the .svg. Sorry. W Nowicki (talk) 22:56, 28 May 2011 (UTC)